Elamur Research⌕
☰
home / papers / 2026 / TRUST: Towards Robust Social Bot Detection via Uncertainty-Guided Pseudo-Labeling and Graph Structure Purification

TRUST: Towards Robust Social Bot Detection via Uncertainty-Guided Pseudo-Labeling and Graph Structure Purification

Paper · 2026 · Ruixuan Xu, Mengting Hu, Zhunheng Wang et al.

Summary

Graph-based social bot detectors usually assume homophily, meaning that linked users tend to share a label. Bots undermine this assumption by deliberately following and interacting with real people. The resulting bot–human (heterophilous) edges carry misleading signals during GNN message passing, so camouflaged bots can end up classified as humans.

TRUST is a plug-in framework that addresses this in three steps:

  1. Uncertainty estimation. It estimates per-node uncertainty with evidential deep learning (a Dirichlet-based output head). Calibration is sharpened by a new error-sensitive uncertainty loss that pushes uncertainty down for correct predictions and up for wrong ones.
  2. Pseudo-labeling. It assigns pseudo-labels to confidently predicted unlabeled nodes. The cut-off is chosen on the validation set to maximise agreement with gold labels.
  3. Graph purification. It removes edges whose endpoints carry conflicting labels, then retrains the GNN on the cleaned graph.

Experiments on three bot-detection benchmarks with six GNN backbones show consistent gains. This suggests the method works as a general-purpose enhancement rather than a single new detector.

Key points

  • Node features combine four modalities:
    • RoBERTa-base embeddings of the profile description
    • aggregated RoBERTa embeddings of the user's tweets
    • normalised numeric statistics, such as follower/following counts and account age
    • three binary flags: protected, verified, and default profile image
  • Pseudo-labels serve only to decide which edges to cut; they are not added as extra training targets.
  • In ablations on TwiBot-20 with an RGT backbone:
    • Removing either the EDL loss or the ESU loss lowers accuracy, F1 and MCC.
    • Dropping the categorical profile features causes the largest drop among the feature groups.
    • Using both pseudo-labeling and purification beats using pseudo-labeling alone.
  • A case study (TwiBot-20, RGT+TRUST) follows bot node 10,843. It is misclassified as human on the original graph but correctly detected after edge purification.
  • Stated limitation: the framework does not yet exploit large language models. Extending it to LLM-enhanced graph architectures is left for future work.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🏢XiaomiBeijing · China14 works · 434 citations🎓Nankai UniversityTianjin · China36 works · 762 citations🎓Tianjin University of Science and TechnologyTianjin · China2 works

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksSocial Sciences · 1,954 worksInformation Systems · 1,730 worksSociology and Political Science · 626 worksPhysics and Astronomy · 52 worksStatistical and Nonlinear Physics · 47 works

Cite

Cite as Ruixuan Xu, Mengting Hu, Zhunheng Wang, Ming Jiang, Rui Ying, Zhen Zhang, Hang Gao, Shuaipeng Liu and Renhong Cheng. 2026. TRUST: Towards Robust Social Bot Detection via Uncertainty-Guided Pseudo-Labeling and Graph Structure Purification. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1575

@inproceedings{xu2026trust,
  title = {{TRUST: Towards Robust Social Bot Detection via Uncertainty-Guided Pseudo-Labeling and Graph Structure Purification}},
  author = {Ruixuan Xu, Mengting Hu, Zhunheng Wang, Ming Jiang, Rui Ying, Zhen Zhang, Hang Gao, Shuaipeng Liu and Renhong Cheng},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1575},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1575}
}

References

33 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Shangbin Feng, Herun Wan, Ningnan Wang, Jundong Li, and Minnan Luo. 2021a. SATAR: A self-supervised approach to twitter account representation learning and its application in bot detection. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management (CIKM), pages 38… find
  2. Takuya Akiba, Shotaro Sano, Toshihiko Yanase, Takeru Ohta, and Masanori Koyama. 2019. Optuna: A next-generation hyperparameter optimization framework. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (SIGKDD), pages 2623–2631.
  3. Shangbin Feng, Herun Wan, Ningnan Wang, Jundong Li, and Minnan Luo. 2021b. Twibot-20: A comprehensive twitter bot detection benchmark. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management (CIKM), pages 4485–4494. find
  4. This work is supported by the National Natural Science Foundation of China (No.62406151). find
  5. Shangbin Feng, Herun Wan, Ningnan Wang, and Minnan Luo. 2021c. Botrgcn: Twitter bot detection with relational graph convolutional networks. In Proceedings of the 2021 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), pages 236–239. find
  6. Shilong Li, Boyu Qiao, Kun Li, Qianqian Lu, Meng Lin, and Wei Zhou. 2023. Multi-modal social bot detection: Learning homophilic and heterophilic connections adaptively. In Proceedings of the 31st ACM International Conference on Multimedia (ACM MM), pages 3908–3916. find
  7. Emilio Ferrara. 2023. Social bot detection in the age of chatgpt: Challenges and opportunities. First Monday, 28(6). find
  8. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692.
  9. Emilio Ferrara, Stefano Cresci, and Luca Luceri. 2020. Misinformation, manipulation, and abuse on social media in the era of COVID-19. Journal of Computational Social Science, 3(2):271–277. find
  10. Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q. Weinberger. 2017. On calibration of modern neural networks. In Proceedings of the 34th International Conference on Machine Learning (PMLR), volume 70, pages 1321–1330. find
  11. Qinglang Guo, Haiyong Xie, Yangyang Li, Wen Ma, and Chao Zhang. 2022. Social bots detection via fusing bert and graph convolutional networks. Symmetry, 14(1). find
  12. William L. Hamilton, Rex Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs. In Advances in Neural Information Processing Systems (NeurIPS), pages 1025–1035.
  13. Kadhim Hayawi, Sujith Mathew, Neethu Venugopal, Mohammad M. Masud, and Pin-Han Ho. 2022. DeeProBot: A hybrid deep neural network model for social bot detection based on user profile data. Social Network Analysis and Mining, 12(1):43. find
  14. Buyun He, Yingguang Yang, Qi Wu, Hao Liu, Renyu Yang, Hao Peng, Xiang Wang, Yong Liao, and Pengyuan Zhou. 2024. Dynamicity-aware social bot detection with dynamic graph transformers. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI), pages 5844–5852… find
  15. Yuhan Liu, Zhaoxuan Tan, Heng Wang, Shangbin Feng, Qinghua Zheng, and Minnan Luo. 2023. BotMoE: Twitter bot detection with community-aware mixtures of modal-specific experts. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), p… find
  16. Qingsong Lv, Ming Ding, Qiang Liu, Yuxiang Chen, Wenzheng Feng, Siming He, Chang Zhou, Jianguo Jiang, Yuxiao Dong, and Jie Tang. 2021. Are we really making much progress? revisiting, benchmarking, and refining heterogeneous graph neural networks. Preprint, arXiv:2112.14936. — [arXiv:2112.14936](https://arxiv.org/abs/2112.14936) find
  17. Nuoyan Lyu, Bingbing Xu, Fangda Guo, and Huawei Shen. 2023. DCGNN: Dual-channel graph neural network for social bot detection. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management (CIKM), pages 4155–4159. find
  18. Lynnette Hui Xian Ng and Kathleen M. Carley. 2023. Botbuster: Multi-platform bot detection using a mixture of experts. In Proceedings of the International AAAI Conference on Web and Social Media (AAAI), pages 686–697. find
  19. Boyu Qiao, Wei Zhou, Kun Li, Shilong Li, and Songlin Hu. 2025. Dispelling the fake: Social bot detection based on edge confidence evaluation. IEEE Transactions on Neural Networks and Learning Systems, 36(4):7302–7315. find
  20. Ziniu Hu, Yuxiao Dong, Kuansan Wang, and Yizhou Sun. 2020. Heterogeneous graph transformer. In Proceedings of The Web Conference (WWW), page 2704–2710. find
  21. Murat Sensoy, Lance Kaplan, and Melih Kandemir. 2018. Evidential deep learning to quantify classification uncertainty. In Advances in Neural Information Processing Systems (NeurIPS), pages 1183–1193. find
  22. Thomas N. Kipf and Max Welling. 2017. Semi-supervised classification with graph convolutional networks. In Proceedings of the International Conference on Learning Representations (ICLR).
  23. Chengcheng Shao, Giovanni Luca Ciampaglia, Onur Varol, Kai-Cheng Yang, Alessandro Flammini, and Filippo Menczer. 2018. The spread of low-credibility content by social bots. Nature communications, 9(1):4787. find
  24. Zhenyu Lei, Herun Wan, Wenqian Zhang, Shangbin Feng, Zilong Chen, Jundong Li, Qinghua Zheng, and Minnan Luo. 2023. BIC: Twitter bot detection with text-graph interaction and semantic consistency. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL), pages… find
  25. Onur Varol, Emilio Ferrara, Clayton A. Davis, Filippo Menczer, and Alessandro Flammini. 2017. Online human-bot interactions: Detection, estimation, and characterization. In Proceedings of the International AAAI Conference on Web and Social Media (AAAI), pages 280–289. find
  26. Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio. 2018. Graph attention networks. In Proceedings of the International Conference on Learning Representations (ICLR). find
  27. Chao Yang, Robert Harkreader, and Guofei Gu. 2013. Empirical evaluation and new design for fighting evolving twitter spammers. IEEE Transactions on Information Forensics and Security, 8(8):1280–1293. find
  28. Kai-Cheng Yang, Onur Varol, Pik-Mai Hui, and Filippo Menczer. 2020. Scalable and generalizable social bot detection through data selection. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), volume 34, pages 1096–1103. find
  29. Yingguang Yang, Qi Wu, Buyun He, Hao Peng, Renyu Yang, Zhifeng Hao, and Yong Liao. 2024. Sebot: Structural entropy guided multi-view contrastive learning for social bot detection. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (SIGKDD), page 3841–3852. find
  30. Sen Ye, Zhaoxuan Tan, Zhenyu Lei, Ruijie He, Hongrui Wang, Qinghua Zheng, and Minnan Luo. 2023. HOFA: Twitter bot detection with homophily-oriented augmentation and frequency adaptive attention. Preprint, arXiv:2306.12870. — [arXiv:2306.12870](https://arxiv.org/abs/2306.12870) find
  31. Kaqian Zeng, Zhao Li, and Xiujuan Wang. 2025. Emoji-driven sentiment analysis for social bot detection with relational graph convolutional networks. Sensors, 25(13). find
  32. Ming Zhou, Dan Zhang, Yuandong Wang, Yangli-Ao Geng, and Jie Tang. 2023. Detecting social bot on the fly using contrastive learning. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management (CIKM), pages 4995– 5001. find
  33. Jiong Zhu, Yujun Yan, Lingxiao Zhao, Mark Heimann, Leman Akoglu, and Danai Koutra. 2020. Beyond homophily in graph neural networks: Current limitations and effective designs. In Advances in Neural Information Processing Systems (NeurIPS), pages 7793–7804. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/xu-2026-trust-robust-social-bot-detection/